why FreeBSD and not Linux. This is a blunder IMO.
As your subsequent posts show, you know about the IP issue with GPL. Because of GPL, WIND cannot integrate their toolset with Linux, or extend Linux to new platforms, multicore or otherwise. What you may not have thought about is that the problem is not unique to WIND. The licensing problem extends to all commercial developers with valuable IP -- like all of WIND's partners.
BSD Unix has a rich history as one of the foundation platforms of Unix today, and it continues to evolve with a free source community just like Linux. The difference is that FreeBSD has a “business-friendly” license permitting commercial companies to protect IP.
When using Linux as a shrink-wrapped OS on a standard hardware platform, the license problem may not surface as it does with embedded systems. It's when you port Linux to a new processor, or integrate VxWorks or Virtuoso with Linux in some exotic, multicore design that the problem arises. Since the exact boundary separating what constitutes altering the Linux code base and what doesn’t is subject to interpretation, there is almost nothing WIND can do with Linux without risking valuable IP. Again, this same barrier applies to most of WIND’s partners also.
One thing all embedded devices have in common, from the smallest DSP chip, 3G wireless phones, to the largest communications/server appliance is that the user generally cares only about the functionality of what is inside. Embedded devices are black boxes, not computers. If two NAS devices stand side by side, function similarly, and are managed indistinguishably, then they are the same, irrespective of OS details. If it looks like a duck, quacks like a duck, it is a duck. In fact, most embedded devices purposely hide the identity of the underlying OS, choosing to brand on the basis of functional specifications. (VxWorks has succeeded beyond its peers, not because it was distinguishable from competition, but because development teams can get reliably to market faster with less risk than possible with the competition.)
Since applications that run on Linux are a mere recompile away from being executable on BSD Unix, the effective difference between Linux and BSD Unix can be discerned only by the truly technically religious. This means that a BSD Unix NAS device easily can be equipped with the same management, security and connectivity software that a similar Linux device might use; thereby making it exactly as saleable as would be an equivalent Linux device. In other words, there are no network effects favoring a Linux-based NAS device from a functionally equivalent BSD Unix device.
The neat thing for WIND about the BSD Unix business model is that a large, highly motivated user community (obviously including computer scientists, as indicated by the professor’s expressed concern on the conference call) maintains FreeBSD, which forms the generic foundation to commercial ports and tweaks of that are licensed by WIND. By owning the freely maintained code base, WIND can provide a soup-to-nuts, integrated solution from the smallest DSP core, to traditional embedded microprocessors, to generic, application-rich computing environments.
Over the last couple of years, I have become convinced vertical solutions are the best means for WIND to secure niche markets with very high ASPs. What you will now see are integrated vertical solutions that weave complex combinations of virtuoso, VxWorks and BSD Unix. ASP will skyrocket because the only thing left for the OEM to do is the form factor and a thin layer of application code (or script), and there will be no competition offering anything close to the same level of integration. Examples abound in the network communications arena, but specifically I would expect outstanding vertical solutions to appear soon in the server appliance space, specifically NAS, SAN and InfiniBand devices.
Allen |
